Pythia8_i CMake Fix, master branch (2022.12.09.)
Cleaned up the CMake configuration of Pythia8_i to fix the issue that @tadej noticed this morning in the Athena and AthGeneration nightlies.
- http://atlas-computing.web.cern.ch/atlas-computing/links/distDirectory/gitwww/MASTERWebArea/ardoc_web_areamaster64BC7G11AthenaOpt/ARDOC_Log_2022-12-08T2101/ardoc_cpack_combined.html
- http://atlas-computing.web.cern.ch/atlas-computing/links/distDirectory/gitwww/MASTERWebArea/ardoc_web_areamaster64BC7G11AthGenerationOpt/ARDOC_Log_2022-12-08T2101/ardoc_cpack_combined.html
The problem stem from the same sort of thing that we cleaned up with @fwinkl a little while ago. That one should not use the "singular version" of the <FOO>_INCLUDE_DIR variables. Only the <FOO>_INCLUDE_DIRS variables should be used. Of course in this package there was a reason why HEPMC3_INCLUDE_DIR was used. So I had to do some not completely trivial CMake trickery to do things correctly.
At the same time made sure that the Pythia8_iLib library would declare all of its other dependencies correctly as well.
Note that this setup is still not perfect. The ${HEPMC3PYTHIA8INTERFACE_INCLUDE_DIRS} path is not added to the ROOT_INCLUDE_PATH environment variable with this setup. (It's a long story.) Neither was it added with the previous setup, so this is not making anything worse. Just keep in mind that if we decide to generate dictionaries that would need that Pythia8ToHepMC3.h header, we'll need to do things in a slightly more "disruptive" way.